A Fully Magnetic Resonance Imaging-Compatible, Nonelectrical System for Near-Real-Time Tremor Monitoring during Magnetic Resonance-Guided Focused Ultrasound Authors: Adrien Lavalley, Orane Lorton, Paul Constanthin, Vanessa Fleury, Shahan Momjian Published: 2026 ABSTRACT AbstractIntroductionTremor quantification during magnetic resonance-guided focused ultrasound (MRgFUS) for essential tremor remains a clinical challenge. Current methods, such as accelerometry, can still be limited by electromagnetic artifacts in certain cases, magnetic resonance imaging (MRI) incompatibility, and indirect correlation with sonication timing. We aimed to develop a fully MRI-compatible, nonelectrical system capable of continuous, electromagnetically artifact-free tremor monitoring with near-real-time frequency analysis, precisely synchronized with ultrasound delivery. The primary purpose of this tool is to assist with targeting during the verification stage by providing the tremor amplitude in situ.MethodsWe designed a pneumatic device using a standard, easily available endotracheal cuff inflated to a fixed pressure, held by the patient during MRgFUS sonications. Tremor-induced pressure fluctuations were transmitted outside the MRI suite and digitized using a custom Arduino-based system. A software tool was implemented to calculate the tremor amplitude in near-real time.ResultsThe pressure system allowed real-time visualization of the tremor and assessment of the tremor reduction at the verification stage of 10 patients. The mean tremor reduction was 86% immediately after the end of the verification sonication, confirming the correct position of the target. All patients had a lesion at this location except one, in which the target was moved because he experienced sensory side effects.ConclusionWe designed and implemented a near-real-time pressure system for quantification of the tremor amplitude during sonications. The inexpensive and MR-compatible device is easy to implement on patients requiring MRgFUS treatments and avoid the possible loss of information between the temporary tremor reduction and the clinical evaluation. Introduction Essential tremor (ET) is the most prevalent movement disorders. It markedly impairs patient’s quality of life, as it predominantly manifests as an action tremor affecting the hands [1, 2]. When pharmacological treatments fail, lesioning procedures targeting structures such as the thalamic ventral intermediate nucleus or the dentato-rubro-thalamic tract are considered, with magnetic resonance-guided focused ultrasound (MRgFUS) emerging as a noninvasive alternative to deep brain stimulation [1–3]. Although MRgFUS allows real-time monitoring and control of lesioning via magnetic resonance imaging (MRI) thermometry, current clinical practice still lacks an easily implementable, objective, MRI-compatible real-time feedback regarding tremor suppression [4, 5]. Intra-procedural evaluation of clinical efficacy relies largely on subjective visual assessment during inter-sonication breaks, which restricts the ability to dynamically optimize targeting and sonication parameters [6]. Drawing-based tasks and spiral tests are simple to administer and can be analyzed quantitatively using metrics such as tremor amplitude and frequency spectrum [7–9]. However, they are performed outside the MR bore and only after sonication, precluding continuous, real-time feedback during treatment delivery. These limitations have spurred a growing interest in quantitative, real-time tremor monitoring technologies, particularly MRI-compatible solutions that can be integrated into the MRgFUS workflow. The development of an objective tremor quantification tool could improve fine targeting and sonication titration by reducing inter- and intra-rater variability inherent to visual clinical assessment. Moreover, a real-time tremor sensor is especially valuable for immediate target validation, as the tremor may come back between the end of the sonication and the beginning of the clinical evaluation. Accelerometry is now the most used tool for real-time or near-real-time tremor monitoring [10–13]. MRI-compatible accelerometers provide direct, high-resolution measurements of tremor amplitude and frequency, allowing quantification during both rest and motor tasks. They show high sensitivity, MRI compatibility, easy integration into workflow, and a strong correlation with clinical outcomes (e.g., clinical rating scale for tremor [CRST] scores [7]). Although accelerometers are now the most practical tool for tremor monitoring during MRgFUS, their implementation still faces several challenges: accurate measurements depend on proper sensor placement [12, 14, 15], real-time data require complex processing to filter noise and interpret clinical significance [11, 16], and accelerometry signals can still be susceptible to artifacts from MRI-related signals [11, 17–19]. In addition, the presence of a metallic device in the MR environment may affect MR thermometry due to B0 drift, B0 or B1 inhomogeneity, and eddy current-related field perturbations if conductive loops exist [20, 21]. A moving metallic device, such as an accelerometer, is even more problematic as it can distort the B0 field and induce short-time scale heterogeneities, which can result in noise or temperature drift. To the best of our knowledge, the precision of the MR thermometry in the presence of accelerometers has not been investigated. Insufficient thermometry precision could lead to lesions of inadequate size or shape or to clinical decisions biased by erroneous thermometry maps, ultimately resulting in inadequate treatment. Though promising, full MRI compatibility remains challenging, and real-time integration into MRgFUS procedures is still under development. As an alternative to accelerometry in the MR environment, we developed a novel, inexpensive, and fully MRI-compatible pneumatic system for near-real-time tremor monitoring, designed using easily available medically approved components, and adapted for seamless integration into the MRgFUS workflow. This tool for visualizing tremor amplitude during the sonication is primarily useful for guiding the procedure at verification stage by confirming without delay the target coordinates in situ and quantifying the tremor reduction. Methods Patient Population Ten patients with medication-refractory ET (5 men and 5 women) were prospectively enrolled between April 2025 and December 2025. The mean age of the patients was 70 ± 8 years, with a condition that has been present for 22 ± 15 years. All patients underwent thalamotomy using the Exablate Neuro system (Exablate 4000, Type 1.1, v7.33, Insightec, Tirat, Israel). Their mean SDR was 0.45 ± 0.06. Seven patients underwent a first side (unilateral) thalamotomy, while three patients underwent a second-side thalamotomy following a prior contralateral procedure. The target was the ventralis intermediate nucleus of the thalamus (Vim). Four patients had right thalamotomy and six patients had left thalamotomy. Tremor data were acquired on the day of the treatment. The study was approved by the local Ethics Committee and all participants provided written informed consent (CCER-ID2025-00828). Clinical Evaluation Tremor was assessed using the clinical scale for tremor (CRST) [22] before the procedure and one day after the procedure. On the day of the procedure, the clinical assessment carried out by the neurologists consisted in evaluating the tremor by asking the patient to draw spirals and lines after each sonication. The tremor amplitude (in mm) was visually estimated in the wing-beating position and compared with the baseline value prior to the first sonication. Side effects were also assessed after each sonication. Table 1.Clinical outcomes of the patients after MRgFUS treatment and technical data from the verification stage​Patient 1Patient 2Patient 3Patient 4Patient 5Patient 6Patient 7Patient 8Patient 9Patient 10Maximum average temperature, °C53.750.348.850.849.150.952.852.852.151.0Sonication duration, s15.012.111.112.212.114.120.013.115.011.0CEM43, min246.013.46.819.16.927.716.449.736.215.9Tremor reduction according to the pressure sensor, %99858693739268829786Tremor reduction according to the wing-beating position, %839040100067100675017Tremor reduction according to CRST at 1 day, %766810072431758957886Side effects 1 day post-MRgFUS​​​​​​​​​​ Sensory––001–00–0 Balance–022––––0– Dysmetria–01–––––0– Paresis2–––1––––0 Dysgeusia0––2–––––– Dysarthria1–1––––––– Fatigue–––2––1–00For side effects, 0, mild; 1, moderate; and 2, severe. Clinical outcomes of the patients after MRgFUS treatment and technical data from the verification stage For side effects, 0, mild; 1, moderate; and 2, severe. Pressure-Sensing System We designed a nonmetallic, nonelectrical, fully MRI-compatible pressure-sensing system using components already approved for MR use. The core sensing element consists of a standard 7.5 ID Medtronic Shiley endotracheal tube (Medtronic, Dublin, Ireland), selected for its compatibility and on-board pneumatic chamber (Fig. 1a). The pneumatic chamber is inflated to a fixed pressure using a calibrated Medtronic Shiley manometer, ensuring repeatable preload conditions across patients. A pencil was inserted in the tube to facilitate the patient’s understanding of how to use and maintain the device. a MR-compatible part of the pressure system composed of 1. endotracheal tube with inflated chamber, 2. tubing, 3. manometer, and 4. pencil. b Picture of a patient holding the pressure sensor during calibration acquisitions. The red arrow indicates the inflated chamber. Before the first sonication, calibration acquisitions were performed to find the position that elicited maximal tremor in the patient’s hand. The patients were instructed to adopt a pen-like posture, allowing tremor-induced grip oscillations to transmit pressure fluctuations (Fig. 1b). These fluctuations are relayed via a standard medical-grade pressure line exiting the MRI suite, connected externally to a clinically approved pressure transducer. The transducer output is fed into an adjustable-gain, high-sensitivity, custom-made amplifier, whose analog signal is digitized using a custom-coded Arduino microcontroller. The system’s sampling rate (29 Hz) and resolution are optimized to detect frequency bands in the 0.1–14 Hz range, corresponding to physiological tremor dynamics. To correlate pressure changes with sonications, we constructed a dedicated electronic circuit that monitors the acoustic output signal of the MRgFUS control software. This circuit, hard-wired to the MRgFUS control computer, detects the audible cue generated during each sonication event and sends a trigger pulse to the Arduino, which timestamps the data stream accordingly. This enables precise temporal alignment of tremor signals with the thermal delivery phase of each MRgFUS pulse, designed to enable dynamic assessment of tremor modulation during sonication. Fast Fourier Transform A Python script (Python 3.12) was developed on an external laptop to visualize the pressure waveforms and the sonication phase on the same graph immediately after the acquisition (see Fig. 2). The tremor data correspond to a tremor in arbitrary units (AUs) measured by the sensor, allowing only relative analysis. A moving window was implemented to display the corresponding fast Fourier transform in a second graph to extract tremor frequency component (Fig. 2). A baseline correction using B-spline smoothing was applied to the tremor amplitudes to correct the low-frequency movements of the patients, occurring, for example, when readjusting the position. a Tremor signal from the pressure sensor over time in patient 3. The sonication event is indicated by the black line with the start and end of the sonication indicated by ON and OFF. The red rectangle indicates the 10s moving windows used to calculate the FFT of the signal. b FFT of the signal contained in the corresponding 10s moving window, before the sonications start. The peak tremor amplitude reaches 1.00 AU at 5.9 Hz after baseline correction. c Tremor signal from the pressure sensor over time in patient 3. The sonication event is indicated by the black line with the start and end of the sonication indicated by ON and OFF. The red rectangle indicates the 10s moving windows used to calculate the FFT of the signal. d FFT of the signal contained in the corresponding 10s moving window, after the sonication. The peak tremor amplitude reaches 0.06 AU at 5.9 Hz after baseline correction. AU, arbitrary unit; FFT, fast Fourier transform. Recording Tremor recording was performed during the verification sonication and included the duration of sonication, as well as the 10 s before and after sonication. Tremor amplitude (in %) was calculated as the ratio between the peak amplitude at the tremor frequency during the 10 s before the sonication and the peak amplitude at the same frequency over time with a margin of ±0.5 Hz (Fig. 3a), allowing the evaluation of the tremor reduction. The average temperature and the thermal dose (cumulated equivalent minute at 43°C [CEM43]) [23] were added to the tremor amplitude graph during offline processing. The averages for all patients regarding tremor amplitude, temperatures, CEM43 values, and their respective 95% confidence intervals are shown in Figure 3b. a Illustration of the tremor amplitude (blue line), temperature (green line), CEM43 (red line), and sonication event with the start and end of the sonication indicated by ON and OFF (black line) over time for patient 4. b Average tremor amplitude (blue line), average temperature (green line), and average CEM43 (red line) with their confidence interval 95% (shaded areas) for all patients over time from the beginning of the sonication. Results Patient Outcome The mean tremor reduction according to the CRST sub-score of the treated hand one day after the treatment was 69.4 ± 19.6%. The mean tremor reduction according to the posture test (wing-beating position) during clinician evaluation following the verification sonication was 61.3 ± 25.0%. Technical information, clinical outcomes and side effects for each patient are detailed in Table 1. Feasibility Near-real-time tremor recordings were successfully obtained in all 10 patients during MRgFUS thalamotomy using the pneumatic pressure-sensing system. Continuous tremor signals could be acquired during sonication without interfering with the treatment workflow. Signal Acquisition Representative tremor signals and corresponding frequency-domain analyses are illustrated in Figure 2 for patient 3. Before sonication, a clear tremor peak was identified at 5.9 Hz, with a normalized amplitude of 1.00 AU after baseline correction. After sonication, tremor amplitude at the same frequency decreased to 0.06 AU, corresponding to a tremor reduction greater than 90%. Tremor amplitudes over time for 1 patient and the average of all patients are shown in Figure 3. Tremor Reduction during Verification Sonications At the verification stage, as summarized in Table 1, the mean average temperature was 51.2 ± 1.3°C (range 48.8–53.7) reached in 13.6 ± 2.0 s, and the mean thermal dose was 43.8 ± 41.6 CEM43 (range 6.8–246.0). The mean tremor reduction measured by the pressure-sensing system at the end of the verification sonication was 86 ± 7% (range 73–99). All patients had a tremor reduction greater than 70% after the verification sonication. Final confirmation of target adequacy and decision to proceed with lesioning were based on standard intra-procedural neurological examination performed after sonication, including assessment of clinical benefit and side effects. In 9 patients, lesioning was subsequently created at this location. In 1 patient (patient 5), the target was moved 0.2 mm anteriorly after the verification sonication because the patient experienced sensory side effects. In some patients, marked tremor reduction was observed at relatively low thermal doses and temperatures. For example, in patient 3, an 86% tremor reduction was observed at a peak temperature of 48.8°C and a thermal dose of 6.8 CEM43 during the verification sonication. Patient Feedback and Tolerability All patients were able to follow the instructions and hold the pen-like pressure-sensing device during sonication. Patients reported subjectively perceiving a marked reduction or complete suppression of tremor during or immediately after sonication, consistent with the recorded pressure signals. Discussion In this study, we developed and reported the feasibility and initial clinical experience of a novel, fully MRI-compatible, pneumatic-based system enabling near-real-time tremor monitoring during MRgFUS thalamotomy for ET. The system allowed continuous acquisition of tremor-related signals during sonication without interfering with the procedural workflow, addressing a key unmet need for intra-procedural, objective tremor assessment in the MR environment. The device, consisting of inexpensive and standard electronic components can be built by anyone with an interest in electronics and a basic electronic skill. However, this device addresses some limitations of existing technologies by enabling the continuous, electromagnetic artifact-free acquisition of tremor signals, with a precise temporal synchronization to ultrasound sonication events. Unlike accelerometry, which relies on electronic components and magnetic-sensitive sensors placed near or inside the MR bore [12, 16], our system uses electrically inert, nonmetallic components, all of which are already approved for use within the MRI environment. This configuration avoids susceptibility to electromagnetic interference and contamination from MR signals on tremor measurements and, above all, from accelerometer wires in the MR bore to MR signals (thermometry), ensuring the reliability and safety of MRgFUS data acquisition. On the other hand, accelerometers requiring careful orientation can suffer from signal drift or clipping, and can still be exposed to distortions in the MRI environment. Additionally, they often require elaborate calibration and filtering pipelines [11, 16]. The pneumatic sensor enabled a rapid and objective assessment of the tremor in near-real time, reducing the risk of potential loss of information between the occurrence of the effects, which are expected to be transient in the verification phase, and the assessment of the tremor. It is able to capture pressure fluctuations induced by tremor through a sealed, pre-inflated endotracheal tube, and to transmit those oscillations mechanically outside the MRI suite, where they are converted to a signal reflecting the tremor. The tremor signal is time-locked to the sonication enabling to analyze the correlation between tremor dynamics and therapeutic ultrasound delivery. Our system uses tubes that are readily available in standard clinical practice and basic electronic circuits that do not represent a significant cost. However, our system does not provide quantitative values of tremor amplitude, but only a quantitative value of tremor reduction, which is eventually the primary endpoint. When comparing the tremor improvement measured by the pressure sensor and the tremor improvement measured by the neurologists, no correlation was found. That may be explained by several reasons in addition to the low number of patients. First, this paper reports only on 10 patients with more than 70% of tremor improvement, which excludes a wide range of possible data that could demonstrate a correlation. Second, the pressure sensor recorded postural tremor while neurological tests mainly included action tremor which are subjectively evaluated (spirals, lines), and the postural test (wing-beating position) performed by the neurologists is visually evaluated with a precision of 10 mm, which is not sufficient to demonstrate a correlation with the sensor. Third, the data processing considered a possible shift in the tremor frequency less than ±0.5 Hz as the shift in the fundamental tremor frequency has been reported to be around –0.16 ± 0.41 Hz [24] before and after each sonication. However, patient 5 had a 73% tremor improvement according to the tremor sensor, and 0% according to the neurologist, because of an unexpected shift of the tremor frequency from 3.5 Hz to 5.5 Hz, visible on the Fourier transform and on the line drawing. Finally, this device was developed to visualize the effect of sonication on tremor before it had time to fade away partially whilst the patient is being moved out of the MRI tunnel. This situation occurred in patient 3, 9 and 10 who had tremor improvement of only 40%, 50% and 17% according to the neurologist, respectively, and 86%, 97% and 86% according to the pressure sensor, respectively. However, following the treatment sonication at this target, their tremor improvement reached 100% in all cases according to the neurologist. In addition, a correlation with the CRST at 1-day post MRgFUS is not guaranteed with low number of patients as the tremor was recorded at verification stage, which does not indicate whether the procedure was successful or not. For example, patient 6 had only 17% of tremor improvement according to CRST at 1-day post-MRgFUS treatment, but the patient interrupted the procedure before the consolidation of the lesion. The treatment workflow was not affected by these recordings, as they take place during sonication, and they did help in the decision-making process. All patients easily understood the instructions and held the pen-like pressure system at the audio signal given by the MR technicians. The system did not prolong or disrupt the treatment workflow. We also noticed an unexpected but welcome effect of real-time visualization of tremor reduction by the patient compared to those with the hand lying on the MR table, particularly for the 3 bilateral treatment patients who could compare to their treatment of the first side (where the sensor was not yet employed). Subjective patient reports of tremor reduction during or immediately after sonication were consistent with the recorded pressure signals, suggesting that real-time feedback may also enhance patient engagement and tolerability during the procedure. In the same way, this position allowed the hand to appear in the field of view of the MRI camera, enabling the team to visualize the tremor reduction in real time and ensure that the patient’s hand was in the correct position in the tunnel. The device may allow to save time in the future, by mainly keeping the evaluation of side effects during the clinical evaluation. We noticed that a significant reduction in tremor can occur at low temperatures, for example, an 86% reduction in tremor at only 48.8°C or 6.8 CEM43 for patient 3. This result suggests that even some targeting sonications during the alignment stage could have a significant impact on tremor. In further study, the pressure sensor could also be used to record the tremor during the alignment sonications to identify the lowest temperature threshold at which a change in tremor can already happen. At this stage, there is no correlation demonstrated between the tremor reduction and the CEM43 delivered. Nevertheless, the present system shows limitations. In certain cases, the baseline inflation pressure of the chamber must be adjusted depending on tremor severity and patient’s grip strength. Insufficient pressure may lead to signal attenuation, while overinflation can reduce sensitivity. This issue limits reproducibility and makes comparisons between sonications difficult. This tool is not applicable to all patients, particularly for those with tremor amplitude <5 mm, patients presenting a broad range of tremor frequencies, or patients mainly presenting shoulder tremor rather than hand or finger tremor. Furthermore, airtight integrity of pressure tubing is essential. Minor leaks or kinks may dampen the signal and distort spectral output. Future iterations of the system could explore the use of noncompressible fluid media, such as hydraulic oil, to improve transmission fidelity, reduce leak risk, and eliminate compliance variability inherent in air-filled chambers. A larger cohort would also be required to strengthen knowledge about the CEM43 or temperature threshold based on tremor reduction. Moreover, the device would require a larger cohort to correlate the tremor reduction measured by the pressure sensor with the tremor reduction assessed by the clinician and validate the device as an intraoperative guidance tool. The 10 patients had a postural tremor sufficient to be detected by the pressure-sensing system, but some patients may only have tremor detectable during movement execution. We can consider asking the patient to virtually draw circles or straight lines with the pen-like device. In conclusion, our system demonstrates a robust, MRI-safe, and clinically feasible solution for near-real-time tremor monitoring during MRgFUS. The system complemented standard clinical evaluation and offers a promising tool for enhancing procedural feedback. All tubing is already available in common practice and does not represent major supplementary expenses. It enables a synchronized acquisition, noise-resistant spectral analysis, and seamless integration into the sonication workflow. With further technical refinement and validation in larger cohorts, it may support a more standardized and responsive decision-making for the treatment of ET, ultimately resulting in improved, safer management of patients. Acknowledgments The authors warmly thank the MR technicians from the radiology department, Yannick Andermatt, Guillaume Metenier, Laurent Koehl, Alain Filograna, and Vincent Breguet, for their help in patient care and giving instructions to the patients. The authors also thank the radiology department for their collaboration on MRgFUS treatments. Statement of Ethics This study protocol was reviewed and approved by the Swiss Ethics Committee “Commission Cantonale d’Ethique de la Recherche sur l’être humain” (CCER), Approval No. CCER-ID2025-00828. All participants provided written informed consent. Conflict of Interest Statement The authors have no conflicts of interest to declare. Funding Sources This study was not supported by any sponsor or funder. Author Contributions A.L. and S.M. designed and built the device. O.L. and S.M. developed code scripts. V.F. was responsible for patient recruitment and clinical evaluation and follow-up. MRgFUS procedures were performed by S.M. and P.C. O.L., S.M., V.F., and P.C. conducted the experiments. O.L. analyzed the results. A.L. and O.L. drafted the manuscript. S.M., V.F., and P.C. reviewed the article. S.M. and V.F. supervised the study.